Back to skills
SKILL.md
Memory Management
ASecurityZero-dependency, filesystem-native 3-Layer Memory Engine based on Google WikiSkill and Stanford graph-planning principles.
- 7 stars
- 0 votes
- 0 copies
- 2 views
- Added September 2, 2026
Security analysis
100/100Pro scans all 15 files and shows the line behind each finding
npx -y skills add richfrem/agent-plugins-skills --skill memory-management --agent claude-codeAre you the author of Memory Management?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/richfrem-memory-management-agent-plugins-skills)---
name: memory-management
plugin: agent-memory
description: Zero-dependency, filesystem-native 3-Layer Memory Engine based on Google WikiSkill and Stanford graph-planning principles.
allowed-tools: Read, Write, Bash
---
# Memory Management (`memory-management`)
Zero-dependency memory system providing high-speed cognitive continuity across agent sessions without external vector databases or daemon processes.
## Contents
- [Critical Constraints](#critical-constraints)
- [3-Layer Architecture](#3-layer-architecture)
- [Quick start](#quick-start)
- [Workflow](#workflow)
- [Verification](#verification)
- [References](#references)
## Critical Constraints
1. **Inference Restriction**: Historical raw execution traces and multi-page wiki dossiers are barred during active task execution to eliminate context window bloat.
2. **Confidence Decay**: Knowledge not re-verified within 30 days decays from `CONFIRMED` to `OBSERVED`.
3. **Asymmetric Persistence Rule**: When an evolution attempt fails, code mutations are rolled back, but wiki insights, edge-case discoveries, and failure logs are NEVER rolled back.
4. **Standard Library Only**: Evolution scripts and core memory lookups must never depend on external database daemons or third-party vector libraries.
## 3-Layer Architecture
- **Layer 1 (Runtime Context)**: Lean `SKILL.md` files (budget <= 100 lines) loaded strictly on-demand.
- **Layer 2 (Compounding Wiki)**: Permanent Markdown in `wiki/` and plugin `references/` (`CONFIRMED`, `OBSERVED`, `HYPOTHESIS`).
- **Layer 3 (Safe Audit Layer)**: Append-only traces in `.agent/learning/traces/cycle_manifests.jsonl` audited via `verify_evolution_receipt.py`.
## Quick start
Execute a native filesystem query across Layer 2 knowledge:
```bash
rg "Status: CONFIRMED" wiki/ references/
```
## Workflow
1. **Session Boot**: Inspect `references/map-debt.md` and relevant domain playbooks in `wiki/` before starting tasks.
2. **Execute & Learn**: Perform task under Layer 1 procedural guidance.
3. **Asymmetric Recording**: On unexpected friction or failure, record findings to `references/map-debt.md` and Layer 2 playbooks before any rollback.
4. **Audit Trace**: Log cycle metadata into Layer 3 audit manifests.
## Verification
Run the memory layer test suite to verify line budgets and native sub-50ms retrieval latency:
```bash
pytest plugins/agent-memory/tests/test_memory_layers.py
```
## References
- [acceptance-criteria.md](references/acceptance-criteria.md) — Acceptance criteria for memory layers and boot sequencing.
- [fallback-tree.md](references/fallback-tree.md) — Fallback protocol when memory files or cache targets are missing.
Files in this skill
- SKILL.md
- acceptance-criteria.md
- assets/resources/memory_architecture.mmd
- assets/resources/memory_lookup_flow.mmd
- assets/resources/memory_session_lifecycle.mmd
- evals/evals.json
- evals/results.tsv
- fallback-tree.md
- references/acceptance-criteria.md
- references/cheapest_models.json
- references/cheapest_models.md
- references/diagrams/architecture/memory_architecture.mmd
- references/diagrams/architecture/memory_lookup_flow.mmd
- references/diagrams/architecture/memory_session_lifecycle.mmd
- references/fallback-tree.md
Attribution
Comments
Loading comments…